FEDERAL UNIVERSITY OF JUIZ DE FORA ENGINEERING DEPARTMENT
ELECTRICAL ENGINEERING GRADUATE PROGRAM
Edgar Bellini Xavier
Economic assessment of an electric vehicle parking lot equipped with photovoltaic generation, energy storage system and electric vehicle chargers
Juiz de Fora 2020
Economic assessment of an electric vehicle parking lot equipped with photovoltaic generation, energy storage system and electric vehicle chargers
Master’s thesis presented to the Electrical Engineering Graduate Program at Federal University of Juiz de Fora, in the area of concentration in Electrical Energy Systems, as a partial requirement to obtain the title of Master in Electrical Engineering.
Advisor: Dr. Bruno Henriques Dias
Co-Advisors: Dr. Bruno Soares Moreira Cesar Borba PhD Jairo Quirós-Tortós
Juiz de Fora 2020
Ficha catalográfica elaborada através do Modelo Latex do CDC da UFJF com os dados fornecidos pelo(a) autor(a)
Xavier, Edgar Bellini.
Economic assessment of an electric vehicle parking lot equipped with photovoltaic generation, energy storage system and electric vehicle chargers / Edgar Bellini Xavier. – 2020.
86 f. : il.
Advisor: Dr. Bruno Henriques Dias
Co-Advisors: Dr. Bruno Soares Moreira Cesar Borba, PhD Jairo Quirós-Tortós
Master’s Thesis – Federal University of Juiz de Fora, Engineering De-partment. Electrical Engineering Graduate Program, 2020.
1. Distribution Networks. 2. Economic Assessment. 3. Electric Vehicles. 4. Parking Lots. 5. Photovoltaic Generation. 6. Real Applications. 7. Storage Systems. I. Dias, Bruno Henriques, orient. II. Borba, Bruno Soares Moreira, coorient. III. Quirós-Tortós, Jairo, coorient. IV. Título.
This work is dedicated to all my family, particularly to my parents, José Eduardo and Maria das Graças, and my sisters Eduarda and Alice. It is also dedicated to my sweet love, Alyne Neves. Without your support, encouragement, and education, nothing would be
possible.
Dedico este trabalho à toda minha família, em especial meus pais, José Eduardo e Maria das Graças, e minhas irmãs Eduarda e Alice. Este trabalho também é dedicado ao meu amor, Alyne Neves. Sem o suporte, incentivo e educação de vocês, nada teria sido possível.
First, I must be thankful to God for giving me strength, wisdom and standing by my side in all moments, especially in the most difficult ones.
I would like to acknowledge my parents, José Eduardo and Maria das Graças. You have always helped me, taught me and support me whenever I needed.
I would also like to thank my love, Alyne Neves, who has always been by my side giving me a pep talk, love and handling my worst and doubtful moments.
To my advisors, Bruno Dias, Jairo Quirós and Bruno Borba, I’m very thankful for all the knowledge you have given me and for the patience you had. I would also like to thank professors Madson Cortes and Leonardo Willer for accepting to be part of the examination board of this thesis and for the contribution to this work.
A special thanks must be expressed to some friends: Daniel Lucena, João Ricardo Pereira, Jean-Philippe Bilger, Gustavo Reis and Lucas Deotti. Each one of you surely has helped me achieve the results in this thesis.
I’m also very thankful to my cousin Paula Bara for her affection, attention, and dedication to helping me edit and correct this thesis.
Finally, I would like to thank the National Council for Scientific and Technological Development (CNPq), the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), FAPEMIG and INERGE for their financial and/or technical support for the
AGRADECIMENTOS
Primeiramente agradeço a Deus por ter me dado força, sabedoria e estado ao meu lado em todos os momentos, principalmente nos mais difíceis.
Também gostaria de agradecer aos meus pais, José Eduardo e Maria das Graças. Vocês sempre me ajudaram, ensinaram e me deram todo o suporte que precisei para chegar até aqui.
Agradeço também ao meu amor, Alyne Neves, por sempre ter estado ao meu lado com uma palavra de conforto, amor e me ajudando nos momentos mais difíceis.
Agradeço imensamente os meus orientadores, Bruno Dias, Jairo Quirós e Bruno Borba. Sou muito grato por todo conhecimento recebido e por toda paciência que tiveram. Agradeço também aos professores Madson Cortes e Leonardo Willer, por terem aceitado serem parte da banca examinadora desta dissertação e pelas contribuições dadas a este trabalho.
Um agradecimento especial deve ser feito para alguns grandes amigos: Daniel Lucena, João Ricardo Pereira, Jean-Philippe Bilger, Gustavo Reis e Lucas Deotti. Com certeza cada um de vocês contribuíram um pouco para os resultados desse trabalho.
Agradeço muito à minha prima Paula Bara pelo carinho, atenção e dedicação em me ajudar com a revisão e correção desta dissertação.
Finalmente, também agradeço ao Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), à Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), à Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) e ao Instituto de Estudos e Gestão Energética (INERGE) pelo suporte financeiro e/ou técnico para desenvolvimento deste trabalho.
ABSTRACT
The electric vehicle (EV) fleet has been growing considerably in the last decades, bringing new challenges and opportunities for the electricity system, especially for the Distribution System Operators. In this regard, it is of the utmost importance that governments adopt policies to ensure a robust infrastructure to serve the electric vehicle owners in order not to discourage them from buying those vehicles. This is important since electric vehicles are an eco-friendly mobility fleet that can reduce fossil fuel dependency, noise pollution, help countries to reach the Paris Agreement’s terms and bring some benefits to the electricity system. In this regard, electric vehicle parking lots (EVPL) can play an important role by providing charging stations to those vehicles. But it is very important that the EVPLs are well located and well sized in order to ensure that they can be profitable. This work presents a methodology to determine the optimal size of an EVPL that will not only charge EVs but also have an energy storage system (ESS) and photovoltaic generation (PV) services. The aim of this thesis is to evaluate the profitability of the EVPL operation for 20 years. The proposed methodology shows that a well-sited and well-sized EVPL can be profitable. Moreover, this thesis shows the importance of an energy storage system (ESS) to ensure the profitability of the EVPL and also the positive impact of the photovoltaic (PV) generation in the EVPL profit, when combined to the ESS.
Keywords: Electric Vehicles. Parking Lots. Photovoltaic Generation. Storage Systems. Economic Assessment. Distribution networks. Real Applications.
A frota de veículos elétricos cresceu consideravelmente nas últimas décadas, trazendo novos desafios e oportunidades para o sistema elétrico, especialmente para as empresas de distribuição de energia. Nesse sentido, é de extrema importância que os governos adotem políticas para garantir uma infraestrutura robusta a fim de atender os donos de veículos elétricos, não os desencorajando a comprar esses veículos. Isso é importante, pois os veículos elétricos são considerados uma frota de mobilidade ecológica que pode reduzir a dependência de combustíveis fósseis, a poluição sonora, ajudar os países a cumprir os termos do Acordo de Paris e trazer benefícios para o sistema elétrico. Nesse sentido, os estacionamentos para veículos elétricos podem desempenhar um papel importante, fornecendo estações de carregamento para esses veículos. Mas é muito importante que esses estacionamentos estejam bem localizados e dimensionados para garantir sua rentabilidade. Esta dissertação apresenta uma metodologia para determinar o tamanho ideal de um estacionamento para veículos elétricos que não apenas os recarregue, mas também seja dotado de um sistema de armazenamento de energia (ESS) e serviços de geração fotovoltaica (PV). O objetivo deste trabalho é avaliar a rentabilidade da operação do estacionamento para veículos elétricos em um horizonte de 20 anos. A metodologia proposta mostra que um estacionamento para veículos elétricos bem locali-zado e dimensionado pode ser rentável. Além disso, este trabalho mostra a importância de um sistema de armazenamento de energia para garantir a rentabilidade do estacionamento para veículos elétricos e também o impacto positivo da geração fotovoltaica no lucro deste estacionamento, quando combinado ao sistema de armazenamento de energia.
Palavras-chave: Veículos Elétricos. Estacionamento. Geração Fotovoltaica. Sistemas de Armazenamento. Avaliação Econômica. Redes de distribuição. Aplicações reais.
LIST OF FIGURES
Graph 1 – Global EV stock 2013 to 2018 . . . 16
Graph 2 – Global EV (2018 to 2030) projected stocks . . . 17
Chart 1 – Summary of the key aspects for sitting and sizing EVPL . . . 21
Board 1 – Examples of use of Real-Data . . . 24
Board 2 – Stochastic and Deterministic Approaches . . . 26
Board 3 – Traffic-flow considerations in some papers . . . 27
Chart 2 – Energy Flow . . . 32
Flowchart 1 – Flow chart to determine the total profits for the considered planning period and each EV penetration level . . . 33
Graph 3 – EV growth - policies comparison for 10% penetration level . . . 41
Graph 4 – Example of EV traffic flow for 5% penetration level . . . 42
Graph 5 – Best ESS battery sizes for each EV penetration level . . . 46
Graph 6 – Best PV system sizes for each EV penetration level . . . 47
Graph 7 – Economic results comparison in a "Strong Policy" scenario . . . 48
Graph 8 – Economic results comparison in a "Slower Policy" scenario . . . 48
Graph 9 – Economic results under a Strong Policy curve - Scenario 2 . . . 50
Graph 10 – Economic results under a Strong Policy curve - Scenario 3 . . . 51
Figure 1 – ESS technologies comparison considering Power Output and Energy consumption . . . 65
Figure 2 – Battery Characteristics Differentiation . . . 66
Figure 3 – Li-ion battery composition and charging-discharging process . . . 69
Board 4 – EV Charging Modes . . . 71
Figure 5 – Charger Modes: A - Mode 2; B - Mode 3; C - Mode 4 . . . 71
Figure 6 – SAE J1772 plug . . . 72
Figure 7 – SAE J1772 pin layout . . . 73
Figure 8 – Mennekes plug . . . 73
Figure 9 – Mennekes pin layout . . . 74
Figure 10 – CHAdeMO plug . . . 75
Figure 11 – CHAdeMO pin layout . . . 76
Figure 12 – CCS Plug . . . 76
Figure 13 – GB/T plug . . . 77
Figure 14 – GB/T pin layout - DC charging . . . 78
Graph 11 – Example of RTP model . . . 80
Table 1 – Inverter costs according to (67) . . . 44 Table 2 – Scenario 1: best configuration for each EV penetration level and financial
results - Strong Policy . . . 45 Table 3 – Scenario 1: best configuration for each EV penetration level and financial
results - Slower Transition . . . 46 Table 4 – Scenario 2: best configuration for each EV penetration level and financial
results - Strong Policy . . . 49 Table 5 – Scenario 3: best configuration for each EV penetration level and financial
results . . . 51 Table 6 – Summary of the NPV results . . . 52 Table 7 – Summary of the IRR (%) results . . . 52
ACRONYMS
ABNT Associação Brasileira de Normas Técnicas ABRASCE Associação Brasileira de Shopping Centers AC Alternate Current
ANEEL Agência Nacional de Energia Elétrica
ANFAVEA Associação Nacional dos Fabricantes de Veículos Automotores BEV Battery Electric Vehicles
CAN Controller Area Network CCS Combined Charging System
CEMIG Companhia de Eletricidade de Minas Gerais CO2 Carbon Dioxide
CS Charging Station
CSP Concentrating Solar Power
DC Direct Current
DG Distribution Generation DSO Distribution System Operator EMS Energy Management Strategy ESS Energy Storage System
EV Electric Vehicles
EVCIPA Electric Vehicle Charging Infrastructure Promotion Agency EVPL Electric Vehicle Parking Lot
EVSE Electric Vehicle Supply Equipment FCLM Flow-Capturing Location Model FCS Fast Charging Station
G2V Grid-to-vehicle
IEC International Electrotechnical Commission IRR Internal Rate of Return
PbA Lead-Acid
LDV Light-duty Vehicles Li-ion Lithium-ion
MILP Mixed Integer Linear Programming MIP Mixed Integer Programming
Na-S Sodium-sulfur Ni-Cd Nickel–Cadmium Ni-MH Nickel–Metal Hydride NPV Net Present Value
O&M Operation and Maintenance
PbA Lead-Acid
PDF Probability Density Function PHEV Plug-in Hybrid Electric Vehicles PSO Particle Swarm Optimization PSB Polysulfide-bromine
PV Photovoltaic
REDOX Reduction-oxidation
RER Renewable Energy Resources RFB Redox Flow Batteries
ROI Return of Investment RTP Real Time Pricing
SAE Society of Automotive Engineers SOC State-of-Charge
T&D Transmission and Distribution
ToU Time-of-Use
UK United Kingdom
UPS Uninterruptible Power Supply V2G Vehicle-to-grid
1 INTRODUCTION . . . . 16
1.1 AIMS AND OBJECTIVES . . . 18
1.2 RELATED PUBLICATIONS . . . 19
1.3 STRUCTURE OF THE THESIS . . . 19
2 BRIEF LITERATURE REVIEW, POSSIBLE FUTURE RE-VENUES AND TECHNICAL ASPECTS . . . . 20
2.1 COMPUTATIONAL EFFORT . . . 21
2.2 STATISTICS . . . 24
2.3 IMPLEMENTABILITY AND SCALABILITY . . . 27
2.4 VEHICLE-TO-GRID AND ANCILLARY SERVICES: POSSIBLE FU-TURE REVENUES FOR EVPL . . . 29
2.5 TECHNICAL ASPECTS . . . 31
2.6 SUMMARY OF THE CHAPTER . . . 31
3 PROPOSED METHODOLOGY . . . . 32
3.1 PROBLEM FORMULATION AND METHODOLOGY PROCEDURE 32 3.1.1 Objective Function . . . 34
3.1.2 Constraints . . . 34
3.1.3 Calculating the planning period profit . . . 36
3.2 SUMMARY OF THE CHAPTER . . . 37
4 RESULTS . . . . 38
4.1 PARAMETERS OF THE ANALYSIS . . . 38
4.1.1 Economic Parameters . . . 38
4.1.2 EV and Chargers Parameters . . . 40
4.1.3 Batteries Parameters . . . 43
4.1.4 PV Parameters . . . 43
4.2 RESULTS . . . 44
4.2.1 Scenario 1 - Battery, PV and chargers available . . . 45
4.2.2 Scenario 2 - Battery only . . . 49
4.2.3 Scenario 3 - PV only . . . 50
4.3 Results Compiled . . . 52
4.4 SUMMARY OF THE CHAPTER . . . 53
5 CONCLUSION . . . . 54
REFERENCES . . . . 57
APPENDIX A – TECHNICAL ASPECTS . . . . 65
APPENDIX B – TARRIFF STRUCTURES . . . . 80
attention to the EV charge and discharge operations, in special the uncoordinated charging and discharging (9) which can hamper the occurrence of the EV benefits.
In this regard, the EVPL might help to avoid the uncoordinated charge and discharge problem since the EVPL operator will be able to define the charge and discharge plans according to the EV fleet and electricity market aspects. However, it is important to bear in mind that the EVPL should not be randomly allocated. If that happens, the EVPL might be located in an inappropriate point of the grid causing problems (i.e.: voltage and frequency fluctuation, increase of peak demand).
To avoid a bad siting of the EVPL, it is of utmost importance that the Distribution System Operator (DSO) provides studies about the integration of this charging infrastruc-ture in order to ensure that the place to install the PL and its size are well suited from the perspective of the EVPL operation and grid impacts.
1.1 AIMS AND OBJECTIVES
The growing deployment of EVs in the last decade shows how important it is for governments to be prepared to provide a robust charging infrastructure in order to increase the EV sales and achieve the benefits of this new transportation technology. In this regard, the Electric Vehicle Parking Lot (EVPL) is a useful resource, but it is important to perform previous studies in order to ensure the profitability and feasibility of the installation.
This thesis provides an economical evaluation of the installation and operation of an EVPL installed in a shopping center or another public strategic place, considering a long term (i.e.: 20 years) operation planning. The economic impact in the EVPL operation is also investigated considering different governments policies that reflect in the growth of the EV fleet. The EVPL may be equipped with an Energy Storage System (ESS) that can be used to store energy to be used to charge the EVs when the energy tariff is higher, Photovoltaic (PV) panels that generate energy to be injected into the grid and also to charge the EVs in the EVPL and/or the battery system and, of course, commercial EV chargers. The economic impact of the installation of an ESS and PV generation will also be investigated.
It is not the aim of this work:
a) To propose an Energy Management System (EMS); b) To propose a charging schedule for EV;
c) To evaluate the impact of batteries on nature;
d) It was not taken into consideration the economic impact of the EVPL providing ancillary services;
19
e) The EVPL was not considered to be providing V2G service.
1.2 RELATED PUBLICATIONS
XAVIER, E. B.; DIAS, B. H.; BORBA, B. S. M. C.; QUIRÓS-TORTÓS, J. Sizing and Placing EV Parking Lots: Challenges Ahead in Real Applications. In: IEEE PES INNOVATIVE SMART GRID TECHNOLOGIES LATIN AMERICA, 2019, Gramado, Brazil. IEEE DOI: 10.1109/ISGT-LA.2019.8895420
XAVIER, E. B.; DIAS, B. H.; BORBA, B. S. M. C.; QUIRÓS-TORTÓS, J. Methodology to Economic Evaluation of an Electric Vehicle Parking Lot Equipped with PV and Storage. In: IEEE PES TRANSMISSION AND DISTRIBUTION CONFERENCE AND EXPOSITION – LATIN AMERICA, 2020, Montevideo, Uruguay.
1.3 STRUCTURE OF THE THESIS
This thesis is organized in six chapters: Chapter 2 presents a literature review about siting and sizing electric vehicle parking lots, and possible revenues that might increase the EVPL profitability, besides that, it also introduces the importance of considering the technical aspects; in Chapter 3, the proposed methodology is detailed; the numeric results to evaluate the proposed methodology is presented in Chapter 4; Chapter 5 presents the conclusion of this thesis and some proposed future works.
Three appendixes are presented to help better understand some concepts: Appendix A details the technical aspects about batteries and electric vehicle chargers; Appendix B summarizes the main tariff structures adopted in some countries; finally, Appendix C presents the Brazilian tariff system.
2 BRIEF LITERATURE REVIEW, POSSIBLE FUTURE REVENUES AND TECHNICAL ASPECTS
In order to properly define a location and the size of an EVPL, there are some key aspects that must be taken into account, as presented by (8). Sometimes not all of those aspects can be considered; the definition of which of them will be considered relies on the level of detail needed and the data availability:
a) Computational Effort: This aspect is related to the proposed method (especially the time needed to achieve the best solution) and the level of details of the model (i.e.: analysis of real data and network data). It is very important that the computational resources to be spent must be according to the objective of the problem. Concerning the modeling adopted, there are two main strategies:
– minimization of costs: the goal is to reduce the global costs of installation and operation (i.e.: land acquisition, municipal fees, market energy costs, maintenance, batteries degradation and so on);
– maximization of the total profit: in this strategy the goal is to maximize the difference between costs and revenues (i.e.: energy and reserve market, parking rates, market interaction with EV‘ owners and so on).
b) Statistics: When trying to define a proper location and size for an EVPL, we are handling with a large amount of stochastic and uncertain data (i.e.: State-of-Charge (SOC), number of EVs and traffic and charging behavior). This is the real nature of EVs and their owners’ behavior. In order to consider those uncertainties, it is extremely important to use statistical methods to define scenarios that can be used in realistic, stochastic detailed studies related to EVPLs, in order to ensure the Return of Investment (ROI) and feasibility of the EVPLs allocation and size. It is very important to say that, when we neglect the stochasticity of the data, we are performing a simplification that does not match the real nature of the EV behavior and owners’ habits;
c) Scalability and Implementability: In this aspect, it is important to consider how the proposed solution will perform in conditions similar to those found in real systems, as well as if the software used can be adopted by the EVPL’s planner and if the solution is scalable. In this regard, it is extremely important to perform tests in conditions (i.e.: distribution systems and scenarios based in real EV observation) similar to those found in the real world.
This section presents previous works which give reasons and support the proposed research. It starts presenting the papers according to the key aspects defined by (8) (as
grid (V2G) mode as can be seen in (12). The aim of this work was to evaluate the impact of the traffic pattern of EVs on the EVPL energy market participation. This evaluation was performed by comparing two EVPL, one that provides only G2V service while the other can provide Grid-to-Vehicle (G2V) and V2G services.
Another strategy that affects the computational effort is to use metaheuristics to achieve a solution. It is important to bear in mind that the use of this optimization strategy is very sensible to the metaheuristic parameters. Therefore, it is extremely important to well define and select those parameters, as they can have significant effect on the model output.
One example of this strategy can be found in (13) in which the authors compare two metaheuristics, Artificial Bee Colony algorithm and the Firefly Optimization algorithm, in order to determine the optimal number of EVPLs that should be allocated in a distribution system. The goal of the objective function of this problem is to minimize costs (i.e.: energy loss of the network, energy imported from the main grid, energy supplied by the DGs of the network and energy supplied by the EVPL during battery discharge to support the network) and maximize the power supplied to the EVPL to charge the EV batteries.
The Genetic Algorithm is another metaheuristic that was considered in some papers, such as (14) and (15). The first proposed a methodology to optimally site and size an EVPL based on direct load control programs of demand response in order to enhance the reliability of the distribution network. The objective function proposed in (14) aims to minimize the investment, maintenance, reliability and energy purchasing costs. The second paper, (15), proposed a method to determine the number of EVPLs and also their capacity and location. The objective function goal is to maximize the profit of the DSO (owner of the EVPL in this case). The profit considered revenues from selling energy to customers and charging electric vehicles in the EVPLs, while the costs considered were the EV discharging costs, installation, and operation and maintenance (O&M).
The Simulated Annealing algorithm was used in (16) to proper locate EVPLs aiming to achieve the maximum profit over a defined planning period, considering the following costs investment for structuring the EVPL, maintenance, batteries aging costs due to V2G and G2V services, charging discount (in order to encourage the use of the EVPL by EV owners), while the incomes considered were from energy market participation and reliability improvement, partnering with the DSO.
A fourth example of metaheuristics that was considered is the Particle Swarm Optimization (PSO) algorithm. In (17) the authors proposed a planning framework to proper site and size different EV charging stations in urban areas from the perspective of a social planner. The PSO was used because the proposed methodology is a NP-hard problem.
23
Besides the use of metaheuristics, some papers also consider the classic optimization methods such as Mixed Integer Programming (MIP) or Mixed Integer Linear Programming (MILP). For example, in (18) the authors proposed a model for adequate location of charging stations based on two main travel behaviors: short and long distance. The goal of the objective function proposed is to maximize the coverage of all EV flows, defining the location of fast and slow chargers. To achieve the solution, the authors used branch-and-cut search to solve a proposed MIP model.
The MILP formulation was used in (19) and (20). In the first paper, the authors proposed a model of the EV power flow due to their traffic flow. Moreover, they analyzed the impact of the EV traffic flow in EVPL and charging station (CS) operation. The proposed objective function, in (19), aimed at maximizing the profit of the aggregator (system player responsible for managing all the EVPL and CS) through selling energy to
EVs and market interactions.
The second paper, (20), proposed a two-stage optimization model in order to allocate EVPLs in distribution systems. In the first stage, the model aimed at determining the optimal behavior of the EVPL, considering the possibility of market interactions (G2V) by the EVPL owner. The goal of the second stage was to proper allocate the EVPL, considering the behavior determined in the first stage and network constraints.
Although the proposed model plays a major role in the computational effort needed to achieve the optimal solution of a model, it is important to have in mind that the use of real data also affects the time needed by the proposed model. Moreover, when a model considers real data, it aims at ensuring that the optimal size and place of the EVPLs is as close as possible to a real application solution (8). Board 1 summarizes some examples of papers that have considered some types of real data and what types these data are. It is important to say that some papers have not considered any real data at all, such as (12, 13, 14, 15). It is noteworthy that not considering real data is an accepted simplification that does not make the results wrong, but only not too close from a real application solution in what concerns the sizing and placing of Electric Vehicle Parking Lots.
Board 1 – Examples of use of Real-Data
Source: Prepared by the author (2020).
2.2 STATISTICS
When trying to determine a proper charging infrastructure which favors the growth of the EV fleet, it is important to have in mind that the data required to be analyzed is vastly uncertain and stochastic. Due to this nature of the EV fleet and owners’ behaviors, it is required to perform statistical analysis in order to determine scenarios that can be used to evaluate the actions that can be performed in order to ensure the Return on Investment (ROI) and feasibility of the EVPLs.
In (15) the authors have considered different Probability Density Function (PDF) to model some of the parameters considered in the problem. A log-normal distribution was considered to define the distance covered by each EV, whereas the arrival time and departure time were modeled by a Gaussian distribution function. Finally, the initial State-of-charge (SOC) of each EV was modeled as a random variable under log-normal PDF.
Another example of statistical analysis found in the literature is what the author have performed in (10). They divided the Parking lot in two cases: the first was Residential Parking Lot (evening and night-time parking) while the second one was a Business Area Parking Lot (day-time parking). For both cases, the authors assumed that the arrival and departures times were normally distributed (the variance was always 1h, while the mean time was different in each case and if the vehicle was arriving or departing).
A lognormal distribution function was used to generate a sort of inputs (i.e.: EV daily traveled distance) for the first stage of the optimization problem proposed in (20). The same paper also considered the Weibull distributions to determine different probabilistic density functions of wind speed in order to define wind generation scenarios. Still in this paper, the authors have considered four energy and reserve prices scenarios,
25
defined as the average price of 90 days for each season.
The Normal Distribution Function was used in (16) to determine the uncertainties considered in the proposed model: hourly number of newly connected/disconnected EVs and the SOC of EV batteries while connecting to EVPL. The authors defined five different scenarios with the probabilities µ-2σ, µ-σ, µ, µ+σ and µ+2σ.
More robust statistical methods were considered in (17) and (11). In the first paper, the authors used the Monte Carlo Simulation method to sample the EV parking, driving and charging behaviors, considering a 15-minute time interval on the simulation. The second paper, on the other hand, considered the Markov Chain Monte Carlo Simulation to create generalized models to EV arrivals and parking duration and also to determine artificial annual scenarios for the solar irradiance (used in the solar generation).
A final example of statistical analysis was found in (21) according to which the authors defined a series of probability density functions (PDFs) based on the charging behavior of 221 real residential EV owners monitored over a year, across the United Kingdom (UK). The PDFs created can be used in realistic, stochastic detailed studies related to EV in order to consider the impacts of the EV owner behavior (i.e.: number of connections per day, initial/final SOC, start charging time for both weekdays and weekends). The biggest advantage of the PDFs defined in this paper is the fact that they were based on observations of EVs instead of internal combustion engine vehicles, what makes the PDF closer to a real EV owners’ behaviors.
When performing a study to proper site and size EVPL, the data that will be handled is in its majority stochastic. When this stochasticity is neglected, it simplifies the problem but does not match to the real nature of the EV characteristics and the owners’ behaviors (9). Going further, remembering that most of the data is naturally stochastic (i.e.: load, EV demand and SOC) in (22) it was demonstrated that a deterministic approach cannot properly determine the frequency of technical problems and their consequences. Moreover, the stochastic approach can be adapted to special EV conditions such as locations and type of consumers. As can be seen in Board 2 the most common approach is the deterministic.
Board 2 – Stochastic and Deterministic Approaches
Source: Prepared by the author(2020)
A very important stochastic data, when sizing and siting EVPLs, is the traffic flow and the EV owners’ behavior (23). The main reason is that these data will define when and where the EV will be charging, how many EVs will be charging at the same time and how long it will last (24). Furthermore, these data will affect directly the profit of the EVPLs (8).
One of the first attempts to solve the location of charging points, considering the traffic flow, was formulated by (25). The authors proposed a formulation based on the flow-capturing location model (FCLM) and previously extended considering the flow-based and node-based demands. The disadvantage of the proposed method relies on the fact that it assumes that a facility located in a certain path will serve all passing vehicles, while it is known that EVs require multi-charging station system in order to accomplish long journeys (18).
In (26) the authors used transportation models to define a transportation network traffic flow, creating an unconstrained traffic assignment model transportation system behavior. The drawback of this model is that it is not based on real systems like the one proposed in (27), based in Western Denmark historic data from January 2006 to December 2007. Good models of EV behavior were found in (21) according to which the authors developed a series of PDFs to define charging behaviors of EV owners, based on the observation of more than 200 EVs for 2 years. As it can be seen in Board 3, the traffic flow is not taken into consideration in most of the studies.
27
Board 3 – Traffic-flow considerations in some papers
Source: Prepared by the author (2020).
It is important to remark that not considering the traffic flow does not invalidate the method, but it makes it only less close to a real application solution. Furthermore, adjusting the charging infrastructure closer to traffic patterns, gathered from the statistical analysis performed, is very important to encourage people to buy EVs. In other words, as showed in (29) it is important to consider the interaction between the transport system and the power system. Another example that reinforces this can be found in (30) that demonstrated the higher request for public charging infrastructure in long drives, while, for short drives EV owners usually charge the vehicles during the evening at home. Moreover, (31) highlights how psychological factors (i.e.: EV costs/benefits, social influence and consumer’s range anxiety) have an enormous influence not only in the drivers’ behaviors, but also the charging behaviors.
2.3 IMPLEMENTABILITY AND SCALABILITY
In what concerns the implementability and scalability of a solution to site and size EVPLs, it is important to answer some questions such as:
a) How will the model perform in conditions similar to those found in real systems? b) About the software used, can it be adopted by the planner of EVPLs?
c) Is it possible to scale the proposed solution?
A simplified system must be used in preliminary studies, in order to easily analyze the performance of the method. But it is really important to test the proposed solution in
more complex systems to properly evaluate its scalability. An interesting example of a simplified case study used in preliminary studies and then tested in a more complex system was found in (26). In this paper, the authors mixed two different networks, transportation network and distribution network, in order to define a planning strategy to place EV charging stations. Firstly, they evaluated the proposed method in a 12-node transportation network highway coupled with the 33-bus IEEE test system and, in a second step, they tested the method in the IEEE 123-bus test system coupled with 4 12-node transportation networks.
The IEEE test systems are a good starting point to evaluate the performance of the methods. Some examples of IEEE test systems found in the literature are the IEEE 13-bus radial distribution system used in (20) and the IEEE 37-bus radial distribution system considered in (19). Other works performed simulation in simplified generic systems, such as the 28-bus system used by (16), the 33-bus radial distribution systems considered in (13) and a modified version found in (14), the 69 node radial system used in (15).
The main drawback of the IEEE test systems is that they generally do not reflect all the challenges faced in real and large systems. In this way, it is important to test those methodologies in order to validate them to be used in more complex systems like the observed in the real world. An example of a real system used to test a methodology was found in (18). The authors considered that the government will deploy battery charging and recharging stations in order to stimulate the use of EVs in the Dalian District (China).
In (17), the authors considered the development planning map of the Longgang District in Shenzhen (China) in order to proper locate charging stations along the district. The area covers about 196km2 with a population of 740,000 inhabitants and an EV population of 16,000 predicted for 2020. The author also considered a dynamic equilibrium of the EVs coming into and out of the area under study, so the charging demands occur only in the district.
According to (32), the simulation environment must be prepared to exchange information to be used in specialized grid impact simulations and optimally evaluate the performance and economic benefits of EV insertion and the EVPLs. Moreover, it is important that the selected software should be able to handle stochastic data, as well as be widely adopted by EVPL planners.
elated to the simulation environment used in some studies, a few of them used the software Matlab®, which was used in (10) and (14). The last also used the Global Optimization Toolbox from Matlab® in order to use the Genetic Algorithm method to achieve an optimal solution for the proposed method.
Another software used was the high-level modeling system for mathematical op-timization, called GAMS. It is designed for modeling and solving linear, nonlinear, and
29
mixed-integer optimization problems. In (11) the authors used the BARON solver to validate the proposed model. The BARON solver can implement deterministic techniques relying in methods for global optimization (33). The CPLEX solver, designed to solve large, difficult problems quickly and with minimal user intervention, was used in (20) and (19). to solve the Mixed Integer Linear Programming (MILP) formulated in each paper.
2.4 VEHICLE-TO-GRID AND ANCILLARY SERVICES: POSSIBLE FUTURE REVE-NUES FOR EVPL
Although this work has not considered the EVPL to operate using the V2G protocol, this can be an interesting strategy in the near future in order to increase the revenue possibilities and then the profitability of the EVPL business.
V2G uses communication protocols to exchange messages between EVs and power grid in order to control and manage the EV loads (batteries) by the EVPL operator or the DSO. The V2G strategy can be classified in unidirectional or bidirectional V2G. In the first one, the communication occurs only to charge the EV batteries, while in the bidirectional V2G the batteries of EVs might be charged and/or discharged.
Both V2G categories might be useful to the system. Unidirectional V2G might help in grid overloading, system instability and voltage drop issues by providing active power supply (34, 35). Bidirectional V2G not only provides active power supply but also reactive power supply. In this way, bidirectional V2G would provide reactive power support, power factor regulation and support for the integration of renewable energy resources (RER) (36).
This strategy becomes stronger with the increase of RER and their intermittency due to the fact that EVs might act as a load (absorbing the excess of generation provided by RER) or a generator (delivering power to the grid when RER are in the low generation scenarios) (3). Other positive impact of the V2G strategy is the reduction of greenhouse gas emissions when this strategy is applied integrated to distributed RER (37).
Despite the benefits of the V2G strategy, the smart grid and V2G technologies are under development and the main challenges are: communication schemes, power interfaces, battery technology (38). Furthermore, other energy storage system schemes have been proved to be more efficient (pumped hydroelectric storage, fly wheel and concentrating solar power (CSP) are among the developed technologies used worldwide). For example, the CSP has 99% efficiency and can store energy further than EV batteries (39, 40).
Although the V2G strategy has a very positive future perspective, it has not matured yet; therefore, it requires more detailed studies relying, specially, on battery lifetime, communication schemes, weak grid dynamics, network protection and reliability, and so on (7). Other V2G challenges are investment costs, especially in hardware and
software infrastructure (36), and the social barriers. This last challenge refers to the growth of EV fleet and the anxiety range of EV owners who tends to ensure a certain SOC in the EV batteries for unpredicted cases (28, 40).
Despite the-above mentioned benefits, the V2G bidirectional strategy causes battery degradation and must avoid social barriers. This social barriers come from the habit that EV owners tend to have of usually charging the battery with the highest SOC level as possible (36).
Furthermore, V2G and ancillary services are intimately connected since the first can be used to provide this type of service. If an EVPL has the capability to allow EVs to perform V2G process, it is possible to offer high market value services to the grid with minimum effect on the EV storage system, such as regulation, spinning reserve, peak power support, power quality (41)-(47). In Brazil the ANEEL’s Regulation nº697/2015 defines the procedures and parameters to provide ancillary services.
Some papers have analyzed the feasibility of the electric vehicles contribution to the grid ancillary services. In (48) it was analyzed the feasibility of the V2G in acillary services considering the French electric vehicle market and the EV production from 2010 to 2013. The authors considered 8 EV scenarios and different commuting behaviors. The main restrictions to the availability of the V2G ancillary services, according to the author, are related to the need of performing depth cycles (around 80% of discharge), which may lead to a degradation of the EV batteries.
Another interesting approach considered a typical case in the Western Danish power system with large wind power production was found in (49). The authors performed simulations considering the use of an aggregated battery storage model in load frequency control in order to analyze the application of V2G systems to provide power regulation. A real application was found in (50), where the authors performed a practical demonstration of the V2G applied to provide real-time frequency regulation from EVs in the PJM power system.
In (51) the authors analyzed the impact of the journey patterns of EVs related to the capability to provide ancillary services. The study considered the traffic data of 349 electric vehicles from across the UK to explore journey patterns, focusing on duration and range. Based on this data the authors identify generic journey patterns for a range of commercial and domestic users. They identified that in the majority of the cases drivers required less than half of the battery capacity to the daily journey. The authors also demonstrated that the commercial and private fleets profiles can provide a limited peak shaving service. On the other hand, there are some opportunities for those vehicles not used primarily for commuting activities.
31
2.5 TECHNICAL ASPECTS
In order to achieve a technical and economical feasible EVPL infrastructure, it is crucially important to define which technologies will be used. For example, what battery technology will be used (e.g.: Lead-acid or Lithium-ion)? This is necessary to ensure that the EVPL will be profitable for owners and able to offer quality service to consumers. In this regard, the EVPL owner must be aware of the technological options available.
Concerning batteries, since they have experienced a significant evolution in the last years, there are great options depending on the objectives and the investment capacity. On the other hand, the charger market is quite new, compared to the batteries market; however, different available options can be found in charger technology.
In Appendix A there is a detailed discussion about the state of the art of batteries and EV chargers. Their characteristics, as well as their technological details, are presented.
2.6 SUMMARY OF THE CHAPTER
This chapter presented the key aspects to site and size EVPLs in real applications as well as its operation with renewable energy available to do so. The main topics approached were related to computational effort, statistics, scalability and implementability and its impact when determining a methodology to be adopted when planning the installation and operation of an EVPL. This chapter also approached the proposed strategies, available in literature, to operate an EVPL with renewable energy. It also deals with the importance of considering the technological aspects related to batteries and charges.
33
EVPL multiplying the operational profit for the number of weeks in a year (52 weeks). During the first year of the planning period, the investment costs are discounted, as well as the energy demand costs. In the following years, the maintenance costs and the energy demand cost are to be discounted. If any equipment requires replacement due to the fact it has reached its lifespan, the replacement cost will be discounted as well at the respective year. Those costs will be discussed later on. Finally, the results of each year are aggregated in order to determine the planning period total profit. Flowchart 1 summarizes the process.
Flowchart 1 – Flow chart to determine the total profits for the considered planning period and each EV penetration level
3.1.1 Objective Function
In order to minimize the operational costs, based on Figure 2 the objective function can be written as Equation 3.1:
min N X h=1 −Ch
sell×Ebat−EVh + Cbuyh ×Egrid−bath + CP V −bath ×EP V −bath +
+(Ch
buy−Csellh ) × Egrid−EVh + CP V −gridh ×EP V −gridh + Csurph ×Esurph −
−Ch sell×EP V −EVh (3.1) Where:
a) N: Number of discretion time intervals; b) h: time interval;
c) Ch
buy and Csellh : to buy from the grid and sell to EVs, respectively;
d) Ch
P V −bat and CP V −gridh : Costs to send energy from the PV to charge the battery or
feed the grid, respectively; e) Ch
surp: Cost of wasted surplus PV generation;
f) Eh
bat−EV, Egrid−EVh and EP V −EVh : Energy used to charge the EVs from battery, grid
and PV, respectively; g) Eh
grid−bat and EP V −bath : Energy used to charge the battery from grid and PV,
respec-tively; h) Eh
P V −grid: Energy injected into the grid from the PV;
i) Eh
surp: Surplus PV energy generated.
3.1.2 Constraints
To ensure the feasibility of the solution it is necessary to consider the following constraints:
a) EV charge constraint (Equation 3.2): To ensure that all the energy needed to charge the EVs will be supplied, through the use of batteries, PV panels or the grid;
N
X
h=1
n Eh
bat−EV + EP V −EVh + Egrid−EVh
o = N X h=1 Eh ChargeEV (3.2)
35
b) ESS battery charge/discharge constraint (Equation 3.3): This constraint represents the energy balance of the battery. Thus, this constraint relates the time interval h − 1 with the current interval h. In this constraint LChBat and LDischBat represents
the losses when charging and discharging the ESS battery, respectively. Also, the amount of energy stored in the ESS battery at the end of interval h and in the previous interval are Eh
bat and Ebath−1, respectively;
N X h=1 Eh
bat+ Ebath−1+ (1 − LDischBat) × Ebat−EVh −(1 − LChBat) × Egrid−bath −
−(1 − LChBat) × EP V −bath = 0 (3.3) c) PV panels constraint (Equation 3.4): To ensure that all the photovoltaic energy generation is used to charge the ESS battery, charge the EVs, injected to the grid or lost; N X h=1 n Eh
P V −bat+ EP V −gridh + EP V −EVh + Esurph
o = N X h=1 EP Vh (3.4)
d) ESS battery charge/discharge rate limits: Equation 3.5 shows the ESS battery charge rate limit, while 3.6 the discharge rate limit constraint. Bsize is the size of the ESS
battery, in kWh;
N
X
h=1
n
Egrid−bath + EP V −bath o≤∆Charge×Bsize (3.5)
N
X
h=1
Eh
bat−EV ≤∆Discharge×Bsize (3.6)
e) Bounds: No decision variable can have a value lower than 0 (zero). Related to the upper limit:
– ESS battery SOC limits: The SOC in each time interval must be equal or greater than 0 (zero) and and cannot be lower than SOCmin or bigger than
SOCmax, which must be predefined and depends on the battery technology
considered;
– PV generation energy flows: All the energy flows from the PV panels in a specific time interval must be equal or greater than 0 (zero) and cannot exceed the PV generation at that time interval;
3.1.3 Calculating the planning period profit
After determining the profit on the operation of the EVPL considering all of the N time intervals the Objective Function result will give the total profit of this operation. The next step is to calculate the total profit for the planning period. In order to do so, as presented in Section 3.1 the profit for each year must be calculated. Equations 3.7, 3.8 and 3.9 summarize this process. It is important to remark that the EVPL’s operation profit was negative because it is minimizing the costs.
a) Profit of the first year of operation (P _Y ear in Equation 3.7):at this stage, it must be discounted from the operational profit (P rofitOper) the energy demand cost
defined as the product of the contracted energy demand (EDem) with the energy
demand tariff (CEDemand) charged to the A4 consumers group. Moreover, the following
equipment investments must be deducted from the operational profit: – Chargers (Ichs): the money invested to buy the EV chargers;
– Battery (Ibat): the cost of the ESS battery based installed in the EVPL;
– PV generation equipment (IP V): This is related to the amount invested to buy
and install the photovoltaic generation system.
P_Y ear = −52 × P rofitOper−Ichs−Ibat−IP V −EDem×CE_Demand×12 (3.7)
b) Profit of all year, excepting the first year and years during which it was required to replace any equipment (P _Y ear in Equation 3.8): This is done similarly to the profit of the first year. The difference is that at this point there is no deduction of the investment in equipment, but only the maintenance costs of them. The energy demand cost is also deducted the same way as in the first year;
– Chargers (Mchs): EV chargers maintenance costs;
– Battery (Mbat): ESS Battery maintenance costs;
– PV generation equipment (MP V): PV panels maintenance costs.
P_Y ear = −52 × P rofitOper−Mchs−Mbat−MP V −EDem×CEDem ×12 (3.8)
c) Profit of years in which any piece of equipment required replacement (P _Y ear in Equation 3.9: The only difference from the previous equations is that when any equipment reaches its lifespan and needs to be replaced by a new one, at this year
37
the investment cost of all the replaced pieces of equipment is deducted instead of the maintenance costs of them. Equation 3.9 is an example of the photovoltaic generation inverters.
P_Y ear = −52×P rofitOper−Mchs−Mbat−MP V−EDem×CEDem×12−IInvts (3.9)
3.2 SUMMARY OF THE CHAPTER
This chapter presented the proposed methodology adopted in this work in order to evaluate the feasibility of the EVPL operation in a long-term planning period. The energy flow considred and the proposed problem were discussed. Furthermore, the objective function and the constraints were also presented and detailed in this chapter, also.
4 RESULTS
This work analyzed the feasibility of installing an EVPL in a shopping center. In order to do so, different EV penetration levels and configurations were simulated (with and without batteries and PV panels) and the following economic aspects were determined to the best configuration in each case:
a) Net Present value (NPV): this economic index represents the difference between the present value of cash inflows and the present value of cash outflows over a period of time;
b) Internal Rate of Return (IRR): it compares the initial investment and the future project expenses with the potential return of this project. The IRR is expressed as percentage and is based on the project’s cash flow. To say if the project is valuable or not, the IRR must be compared with the investor’s hurdle rate.
The option to consider these two aspects is related to the fact that not always a single economic aspect is enough to determine if investments in a specific project are valuable from the economic perspective. Sometimes, by analyzing only the NPV, for example, it is possible to find positive values, but the IRR is lower than the hurdle rate. In this case, it might lead to the conclusion that investments in such a project are not so attractive.
4.1 PARAMETERS OF THE ANALYSIS
To proper evaluate the proposed methodology, some parameters were assumed. The key parameters are related to the economic aspects (i.e.: dollar exchange rate, planning period, energy and demand tariffs), the EVs and charger characteristics, battery technology and the PV panels details. Although some of them have been cited before in this thesis, they will be summarized in this section.
4.1.1 Economic Parameters
Before the installation of any business, it is very important to perform an economic evaluation of it. This will help investors to have a clearer idea about the return they can achieve before investing their money on this business. In other words, the economic analysis will show the profitability of the investment.
Power system projects normally have a quite long lifespan (15 years or more). Also, the operational costs (i.e.: fuel) of those projects occur after they have been commissioned. So, this expenditure will happen throughout the project lifespan. Therefore, it is of the
39
utmost importance to consider the value of money over time, considering an adequate discount rate (52).
In the present work, a set of real data has been considered to proper evaluate the methodology. The energy costs were obtained from tariffs of a DSO in the southeast of Brazil: Companhia de Eletricidade de Minas Gerais (CEMIG).
a) Planning period: 20 years; b) Discount Rate: 10% yearly;
c) Hurdle Rate: 10% yearly;
d) Dollar exchange rate: 4.00 R$/US$ (53);
e) Contract Demand tariff: 13.95 R$/kW (54). This cost was considered since the EVPL will be a consumer of the A4 group of CEMIG and the tariff considered in this study is the Green Tariff. More details can be found in Appendix C;
f) Tariff to sell energy to EVs: 0.62833 R$/kW h (54);. In order to encourage EV owners to charge their vehicles in the EVPL, the tariff for them to charge their EVs was considered the same as the residential tariff for CEMIG’s group B3 residential consumer.
In this study, the EVPL was considered a heavy load supplied above 2.3kV. In this regard it is necessary to consider that the EVPL will be under the time-of-use (ToU) tariff system. More details about this can be found in Appendix C. So much so, the considered costs from Equation 3.1, presented in Section 3.1.1, are:
a) Ch
buy: due to consumers’ characteristics (i.e.: Shopping Center), in this work the cost
to buy energy from the grid was considered the CEMIG’s Green Tariff for consumers group A4 (2.3kV to 25kV voltage supply). Moreover, it was considered the green flag standard in this work. In this regard:
– Peak tariff (5pm to 8pm): 1.59969 R$/kWh (CEMIG – Green Tariff for consu-mers of the A4 group) (54);
– Off-peak tariff (0am to 17pm and 20pm to 24pm): 0.35666 R$/kWh (CEMIG – Green Tariff for consumers of the A4 group) (54).
b) Ch
sell: in this work it was considered the CEMIG residential green flag tariff (0.62833
R$/kWh) (54) as the price of selling energy to charge the EVs; c) Ch
P V −bat: 0.0001 R$/kWh. To incentivize the use of PV generation to charge batteries,
d) Ch
P V −grid: 0.0001 R$/kWh. In order to incentivize the injection of PV generation
into the grid in order to reduce the monthly net energy consume, this cost was considerably lower than the others;
e) Ch
surp: 100 R$/kWh. in order to penalize the surplus of generation this value was
considerably higher than the other costs.
It is important to highlight that the costs Ch
P V −bat, CP V −gridh and Csurph were
discounted when calculating the weekly profit of the EVPL, since those costs were adopted in order to penalize or encourage some energy flow.
4.1.2 EV and Chargers Parameters
Since it is quite difficult to get data about EVs, specially owners’ behavior, traffic flow was assumed to be the arriving vehicles in a shopping center parking lot during a typical week. The considered shopping center is also located in the southeast of Brazil, in a region with social-economic conditions similar to the region of the DSO. It is important to remark that detailed data about the shopping center parking lot (e.g.: number of vehicles and parking behavior of users) are quite difficult to collect because these are strategic information for this business.
In this regard, this work considered the total vehicles traffic flow presented in (55), that estimated the hourly flow of vehicles in a shopping center parking lot for one week. In (55) the authors collected data from a shopping center located in Rio de Janeiro among several months. The data were analyzed, and it was observed that the parking behavior varies in special days (e.g.: weekends and holidays). From this analysis, the authors defined a typical week in order to avoid outliers and future under/over analysis, and to have an adequate representation of the traffic profile of a generic shopping center parking lot. Based on this, this work considered the EV traffic flow based on the typical week proposed in (55) and the operational results for this typical week were exploited for the year.
In order to define the growth of the penetration level two curves were defined to also evaluate the impact of the EV penetration level growth. To do so, the Brazilian market growth of light vehicles was estimated considering the mean growth of licensed vehicles from 2000 up to 2019 (4% yearly), according to the Brazilian Automotive Industry Association (ANFAVEA) (56). Following, in order to estimate the EV market share growth, two policy-based curves were considered based on (57). The first curve considers a Strong Policy that incentivizes the insertion of EVs (e.g.: tax reduction, dedicated parking spaces) while the second considers a Slower Transition representing either a weaker policy or greater practical or economic obstacles. Graph 3 shows an example of these curves for the 10% penetration growth.
Graph 4 – Example of EV traffic flow for 5% penetration level
Source: Prepared by the author (2020).
Concerning the EVs, since the Nissan Leaf has one the biggest market share worldwide, aspects of this vehicle were considered in this work. In addition, EV chargers will always be considered in the analyses since the main goal of the EVPL is to provide energy to charge EVs. In this regard, the parameters related to the EVs and chargers are summarized below:
a) EV battery size: 40 kW h. In this thesis, it was assumed that all EVs present a battery similar to the Nissan Leaf (58);
b) EV charging rate: 7.4kW h. All the chargers were assumed to be a "DARK Wallbox Tipo 2 32 Amperios – 230V – Manguera"from (59). The EV charging losses were considered in the EVPL batteries losses;
c) Chargers investment costs (Ichs): US$ 860.00 per unit (59);
d) Chargers maintenance costs (Mchs): 2.7% of the investment cost (60);
e) Number of chargers: Varied accordingly to the penetration level; f) Charger life cycle: 20 years.
Since no charging control scheme was considered until this part of the thesis, it was assumed that each EV will be charged for 1 hour and then it leaves the shopping center EVPL. This is in line with a market research made by the Brazilian Association of Shopping Centers (ABRASCE) in 2016 (61). This market research reached the conclusion
43
that consumers stay in a shopping center for 76 minutes in average. Thus, considering 1 hour seems to be a reasonable choice.
4.1.3 Batteries Parameters
As presented in Chapter A there are several battery technologies that can be used as Energy Storage System (ESS) in an EVPL. In this work, the chosen technology was the Li-ion. And the parameters of batteries are as follows:
a) Minimum and Maximum State Of Charge (SOCmin and SOCmax): 25% and 90%,
respectively (13). To avoid damaging the battery during charge and discharge, it is not recommended to charge the battery up to 100% and neither discharge it until 0% (the called deep cycle);
b) Initial State of Charge: 25%, the same as SOCmin;
c) Discharge and Charge battery losses (LDischBat and LChBat): 10% (62);
d) Charge and Discharge rates (∆Charge and ∆Discharge): 100% each (62);
e) Battery investment costs (Ibat): 200.00 US$/kW h (63), considering the installation
costs;
f) Battery maintenance costs (Mbat): 2.7% of the investment cost (60);
g) Battery maximum size: 1500 kW h; h) Batteries life cycle: 20 years.
4.1.4 PV Parameters
The PV generation was installed in order to provide energy to charge EVs, charge the EVPL’s battery or even inject energy into the grid. It is important to remark that in Brazil there is no feed-in tariff; the regulatory agency, Agência Nacional de Energia Elétrica (ANEEL), determined in Res.482/2012 (64) that the distribution generation is based on the net metering concept. Therefore, in the Brazilian energy system the energy generated by the PV panels can only be injected into the grid in order to reduce the energy consumption, and no money is payed to this energy. In this work, the net metering concept was not implemented. In this regard, PV panels are used just to reduce the energy required from the grid by charging EVs and batteries with the energy generated by the PV panels instead of the energy from the grid.
Since the PV generation system (PV panels, inverters, cables and so on) are quite expensive, but with a considerably long-life cycle, it is very likely that it must be well
sized in order not to spend a lot of money in a system that will generate too much energy that it will be lost (neither injected or used in the EVPL) or even invest in a system that will not help to reduce the energy bill since the PV generation is undersized. Following are the parameters of the PV generation system adopted in this work:
a) PV Investment costs considering the inverters and the installation costs (IP V):
1000.00 US$/kW p mean cost from the presented in (65); b) Maintenance cost (MP V): 1% of the investment cost (66);
c) PV generation losses: 0%; d) PV panels life cycle: 20 years;
e) Inverters replacement cost (IInvts): Table 1 summarizes the retail costs of each
inverter module market available according to (67). To simplify the analysis, the installation or taxes costs were not considered. Furthermore, the replacement cost is calculated depending on the PV generation system size;
Table 1 – Inverter costs
kWp R$ 1 2,250.00 2 2,500.00 3 4,500.00 4 5,000.00 5 6,000.00 6 8,500.00 8 10,500.00 15 14,000.00 25 16,500.00 75 37,000.00 Source: (67).
f) Inverters life cycle: 12 years;
g) PV system maximum size: 1500 kW p.
4.2 RESULTS
To evaluate the economic impact of the PV and ESS battery system in the Electric Vehicle Parking Lot (EVPL) operation, 4 (four) scenarios were simulated. Since the considered EVPL aims at charging EVs, in all of the following scenarios the availability of EV chargers was considered:
45
a) Battery, PV and chargers available (Scenario 1);
b) Battery and chargers only (Scenario 2), no PV available; c) PV and chargers only (Scenario 3), no battery available; d) Chargers only (Scenario 4), no PV or battery available.
For Scenario 1, the results for the Strong Policy and Slower Transition scenarios related to the EV penetration level growth were compared in order to evaluate how it can impact the economic results and, therefore, have some influence in the investment decision of possible investors. For Scenarios 2, 3 and 4 it was considered only the "Strong Policy"curve.
4.2.1 Scenario 1 - Battery, PV and chargers available
In this scenario the EVPL operator can use all the equipment (Battery, chargers and PV generation). In this thesis, as presented in Subsection 4.1.2, the penetration level varied from 1% to 10% of the total number of vehicles, considering the Strong Policy and Slower Transition scenarios for EV penetration level growth. Tables 2 and 3 summarize the best configuration (number of chargers units, ESS battery size, PV generation system size) for each penetration level in each scenario of penetration level growth.
Table 2 – Scenario 1: best configuration for each EV penetration level and financial results -Strong Policy
49
it is expected that the IRR for the 5% EV penetration level is bigger than the IRR for the 6% EV penetration level. Despite that, the NPV in both "Strong Policy"and "Slower Transition"scenarios tends to grow as the penetration level increases, but not linearly.
4.2.2 Scenario 2 - Battery only
In this scenario, the PV generation system is not installed. Therefore, the grid is the only way to charge the EVPL’s ESS battery while both the grid and the batteries can be used to charge the EVs. Similar to Scenario 1, the penetration level varied from 1% to 10%, but it was only considered the "Strong Policy"curve. Table 4 summarizes the best configuration in each penetration level.
Table 4 – Scenario 2: best configuration for each EV penetration level and financial results -Strong Policy
Source: Prepared by the author (2020).
In this scenario, the initial investment is considerably low due to the high PV system cost, that is considered in Scenario 1. And this reflects on the IRR, since it can increase up to 4% when compared to the 6% penetration level case in Scenario 1, for example. Graph 9 shows the IRR for each penetration level.
On the other hand, this does not ensure that without PV generation the EVPL will be more profitable. As it can also be seen in Graph 9 the NPV in Scenario 2 is at least 2 times lower when compared to Scenario 1 (considering the same policy for the EV penetration level growth). Moreover, in some cases (i.e.: for 10% penetration) the NPV in the second scenario is lower than the result achieved under the "Slower Policy". This implies that, despite the operation of the EVPL it can be profitable even when
4.2.4 Scenario 4 - Chargers only
Scenario 4 considers the operation of an EVPL without PV system and ESS battery, only the chargers were installed and only the grid can be used to charge the EVs. In this case, it is not suitable to invest in an Electric Vehicle Parking Lot (EVPL), in any penetration level, since the operation will not be profitable. This result shows the importance of investing in PV generation and a storage system when working with electrical vehicle parking lots.
4.3 Results Compiled
Tables 6 and 7, where the black spaces indicate that for this penetration level there was no viable configuration, summarize the main results in each scenario and penetration level.
Table 6 – Summary of the NPV results
53
Table 7 – Summary of the IRR (%) results
Source: Prepared by the author (2020).
4.4 SUMMARY OF THE CHAPTER
This chapter begins defining how the economic aspects will be evaluated and which parameters were considered in this study. Then, it presented the four scenarios that were considered in order to analyze the economic impact of PV generation, batteries and chargers in the EVPL operation. Furthermore, comparison between two policies that reflects the EV penetration growth was presented.
5 CONCLUSION
The Electric Vehicle Parking Lot (EVPL) might be a useful resource for governments that want to increase the penetration of electric vehicles in the country’s fleet. This new fleet will bring great benefits to society (i.e.: less noise in the streets, reduction of urban pollution, reduction of the fossil fuel dependency). Moreover, if EVPLs adopt photovoltaic (PV) generation and an energy storage system (i.e.: batteries) the environmental benefits will increase; besides that, this generation can increase the benefits of the DG to the distribution system.
In order to achieve these benefits and ensure that EVPL operators will be able to have adequate profit, it is necessary to perform proper economic studies to determine the best configuration in each case, since it depends on several factors such as:
a) EV traffic flow characteristics; b) EV owners charging behavior;
c) The Infrastructure technology of choice; d) Location of the EVPL;
e) EV penetration level; f) Energy costs.
Based on this, the work presented a methodology to determine the best configuration of an EVPL and to help in the planning of those EVPLs by governments and investors, considering a study case. Besides that, this methodology is based on the economic evaluation, considering the analysis of the return on investment of those structures. To do so, it considers two different scenarios, the "Slower Policies"and "Strong Policies"EV penetration scenarios.
The first observation is that the EV penetration level growth, on its own, is not a factor that will ensure an earlier payback of the investment made. It is very important to consider the policies adopted in order to estimate how they will interfere in the growth of the EV penetration level and then in the return of the investment. Furthermore, this work showed that there is no linear correlation between the EV penetration level growth and the EVPL Interest Rate of Return (IRR) nor its Net Present Value (NPV).
It was also observed that the investment in ESS batteries is a key aspect to make the EVPL profitable, since without this infrastructure, even in the best case the NPV and the IRR are lower when compared to the scenario with less government incentives to increase the EV penetration level.
